Landscape of surfaceome and endocytome in human glioma is divergent and depends on cellular spatial organization

We set out to develop a platform for tumor surfaceome and endocytome mapping (TS-MAP) that was compatible with human primary stem cell like GBM cells grown in 2D or as 3D-spheroid cultures. As outlined in Fig. 1A, the TS-MAP strategy was subsequently adopted for mouse GBM in vivo models and patient GBM tumors. We initially used primary GBM cells (U3065 and U3082) from the human GBM cell culture (human glioblastoma cell culture [HGCC]) resource (14) to perform comparative quantification of the global surfaceome and the internalized surfaceome fraction in 2D vs. 3D cultures (Fig. 1B). We could show that surface biotinylation efficiency in intact 3D spheroids on a per-cell basis was comparable to 2D cultures, and that ∼50% of the surfaceome was internalized in both culture formats over a period of 90 min (Fig. 1C). The biotinylation reagent used (EZ-Link Sulfo-N-hydroxysuccinimide-SS-biotin) specifically conjugates extracellularly exposed primary amines or amino termini of polypeptides of surface proteins. Its size and negative charge, provided by the sulfonated group, prevents direct cell membrane interaction, e.g., with neutral lipids and unspecific permeation. This was supported by imaging studies, showing comprehensive cell-surface labeling and distinct vesicular structures from internalized proteins in 2D cultures as well as in spheroids (Fig. 1D and SI Appendix, Fig. S1A). Importantly, treatment with MesNa, a membrane-impermeable reducing agent, efficiently abrogated residual cell-surface biotinylation in 2D and 3D cultures, as quantified by fluorescence-activated cell sorting (FACS) (Fig. 1B) and visualized by imaging and Western blotting (SI Appendix, Fig. S1 B and C). We could conclude that the internalized protein signal was not derived from any remaining surface-biotinylated proteins. To corroborate that the occurrence of intracellular biotinylation signal was associated with an endocytic process, GBM cell cultures were costained for biotin and the early endosome antigen 1 (EEA1), showing a clear colocalization in both 2D and 3D models (Fig. 1E). We employed GBM cells expressing the endolysosomal marker CD63 fused with mCherry and live high-resolution confocal microscopy to track translocation of biotinylated surface proteins toward intracellular, CD63-positive vesicles (Fig. 1F and Movie S1). Moreover, using fluorescently labeled ligands of macropinocytosis-dependent and membrane raft-mediated endocytosis (dextran and cholera toxin subunit B, respectively), we confirmed that intracellular biotinylation is the result of endocytic internalization of the surfaceome (SI Appendix, Fig. S1D).

We next built an in-house surfaceome identification (ID) classifier (Fig. 2A) by concatenating single and multipass transmembrane proteins, as well as glycosyl-phosphatidyl-insositol (GPI)-anchored proteins from reviewed (Swiss-Prot) and manually annotated hits using the terms “cell membrane” and “extracellular domain” in UniProt (https://www.uniprot.org/). Additional hits were obtained from gene ontology (GO; http://geneontology.org/) terms “plasma membrane,” “cell surface,” and “external side of plasma membrane.” Retrieved protein IDs were then matched against the SURFY predictor, i.e., a machine learning–based predictor that excludes membrane proteins integrated on the cytoplasmic side lacking an extracellularly exposed motif (http://wlab.ethz.ch/surfaceome) (1). As the reported accuracy of SURFY was 93.5%, the entire SURFY profile (n = 2,886 proteins) was further considered as true hits. Importantly, we identified additional proteins (n = 431), not included in SURFY, by in-depth interrogation of UniProt, GeneCards, GO, and PubMed. We finally curated a total catalog of 3,317 proteins, hereafter referred to as SURFME (available in Dataset S1), which corresponds to ∼16% of the predicted human proteome. SURFME included GPI-anchored (n = 140), α-helical single-pass transmembrane (n = 1,316), and multipass transmembrane (n = 1,750) proteins as well as a small subset (n = 116) of proteins (∼3.5% of SURFME IDs) that did not fit into the major categories (SI Appendix, Fig. S2A).

We next applied the SURFME as a “geneset” within a gene set enrichment analysis (GSEA). In support of the TS-MAP strategy, “surfaceome” and “endocytome” proteins identified in GBM cells displayed a strong enrichment for SURFME when compared to control, nonbiotinylated samples (Fig. 2B). Moreover, SURFME filtering revealed a potential overrepresentation of single-pass transmembrane proteins among experimentally identified surfaceome proteins (∼62% of total identities) vs. the theoretical SURFME set (39.5% of total identities) (SI Appendix, Fig. S2B). Together, we conclude that TS-MAP permits specific enrichment and detection of internalized, low abundant membrane proteins by reducing the contribution of abundant cytosolic and nuclear proteins.

Establishment of the TS-MAP procedure in a 3D-spheroid model allowed us to investigate how the global surfaceome and endocytome is dictated by tumor cell architecture. We found a remarkable divergence in the relative abundance of surfaceome and endocytome proteins in 3D as compared with 2D models. Interestingly, primary glioma cell organization into spheroids was associated with increased internalization of a dominating pool of cell-surface proteins, as shown by differentially expressed protein (DEP) analysis (Fig. 2C). Among SURFME proteins up-regulated in 3D vs. 2D (fold-change [FC] cutoff = 0.5 log₂), 27 were exclusively up-regulated at the surface, 81 were up-regulated at the surface and had increased internalization, and 164 were more endocytosed (Fig. 2D and Dataset S2). In general, up-regulated proteins, both surface resident and endocytosed, were mostly classified as single-pass transmembranes (SI Appendix, Fig. S2C). We found induced expression of several members of the integrin (e.g., ITGAV, ITGB5, ITGA7, ITGB4, and ITGA6), proteoglycan (e.g., BCAN, CSPG5, and GPC1), and semaphorin (SEMA5A and SEMA6D) families in the surfaceome and/or endocytome of 3D as compared with 2D cultures. Accordingly, among overrepresented pathways in spheroids, we identified, e.g., “cell adhesion molecules,” “developmental biology,” “extracellular matrix organization,” and “semaphorin interactions,” which reflect cell–cell and cell–matrix interactions expected to be involved in 3D cellular organization (Fig. 2E). Also, consistent with spheroid formation, we found induction of hypoxia- and acidosis-regulated proteins (SLC16A1/MCT1 and SLC4A4). Interestingly, several proteins associated with cancer stem cells, including in glioma [CD166 (15), F2R/PAR1 (16), EDNRB (17), CX3CL1 (18), ITGA6 (19), and ITGA7 (20)] displayed induced surfaceome or endocytome expression in 3D vs. 2D cultures.

We next selected some of the identified candidate hits, according to 3D vs. 2D differential expression of ≥0.5 log₂ FC or ≤ −0.5 log₂ FC in surface and endocytosed datasets, for immunofluorescence validation (Fig. 2 F–H). In concordance with the TS-MAP data, we found EGFR overexpression in 3D as compared with 2D cultures (Fig. 2 I, Left), and could observe intracellular colocalization of EGFR with endocytosed proteins in spheroids, while this appeared relatively scarce in 2D cultures (Fig. 2 I, Right). Similar results were obtained with CSPG5 (SI Appendix, Fig. S2D), i.e., another candidate from TS-MAP analysis (Fig. 2 F–H). We could further corroborate the down-regulation of CD44 in 3D vs. 2D cultures (SI Appendix, Fig. S2E).

A relevant ADC target should preferentially be highly expressed at the cell surface but also efficiently endocytosed to deliver the cytotoxic payload. To understand whether the increased surface availability and constitutive endocytosis of EGFR in spheroids translated into an increased ADC effect, we next treated 2D and 3D glioma cultures with an anti-EGFR antibody precomplexed with a secondary antibody-monomethyl auristatin F toxin conjugate. We employed an Incucyte imaging system to dynamically follow ADC-induced cytotoxicity in 2D and 3D models over time (SI Appendix, Fig. S2 F and G). In support of increased EGFR-mediated ADC delivery in spheroids, we observed an estimated half maximal effective concentration (EC₅₀) of 0.48 and 2.45 nM in the 3D and 2D cultures, respectively (Fig. 2 J and K).

Together, our results provide a complex understanding of how the surfaceome and endocytome are transformed during spheroid formation, a process generally linked to cancer cell stemness (21). Moreover, validation stainings and ADC treatment studies establish TS-MAP as a high throughput platform for exploration of potential target antigens in vitro.

Although TS-MAP compatibility with intact spheroids is a significant advancement over conventional 2D models, in vitro models fail to fully reflect the complexity of the in vivo situation. To expand on the utility of the TS-MAP approach in vivo, we initially employed a syngeneic mouse GBM model (GL261) known to recapitulate many of the histological features of human GBM (22). GL261 cells were orthotopically injected into the right hemisphere, and following 33 d of incubation, freshly resected tumors as well as nontumor brain tissue from the contralateral hemisphere were processed through the TS-MAP and SURFME platform (SI Appendix, Fig. S3A). We found efficient and comprehensive surface biotinylation of intact tumor tissue sections (SI Appendix, Fig. S3B), as well as in single-cell suspensions (SCSs) from disintegrated tumor and normal mouse brain (SI Appendix, Fig. S3 C and D). SURFME filtering and DEP analysis of liquid chromatography with tandem mass spectrometry (LC-MS/MS) data consistently showed an expected divergence in surface protein abundances between tumor and normal brain (SI Appendix, Fig. S3E and Dataset S3). Out of 346 identified SURFME proteins, 117 were up-regulated in tumor tissue when compared to normal brain, comprising mostly single-pass transmembrane proteins (SI Appendix, Fig. S3F). Similar pathways were overrepresented in tumor vs. normal brain (SI Appendix, Fig. S3G) as found in 3D vs. 2D (Fig. 2E), and included additional tumoral processes, such as “PI3K-Akt signaling pathway” and “integrin cell surface interactions.” Interestingly, GPNMB, the most up-regulated SURFME protein in tumor vs. normal brain (SI Appendix, Fig. S3H) was previously identified as a tumor-associated surface antigen in human GBM (23), and a targeting ADC (glembatumumab vedotin) has entered clinical trials with several tumor types (24). Conversely, the most down-regulated protein, HCN1 (SI Appendix, Fig. S3H), is known as a pacemaker channel widely expressed in the mouse and human brain (25). We conclude that TS-MAP combined with SURFME filtering is compatible with ex vivo surfaceome profiling of intact GBM tumors.

We next adopted the TS-MAP platform to interrogate the surfaceome landscape in a pilot cohort (n = 10) of patient gliomas covering different WHO grades and histologies (Fig. 3A; patient characteristics are presented in SI Appendix, Table S1). Following routine MRI examination, and in parallel with pathological-anatomical diagnosis (PAD), fresh tumor specimens were dissected into 0.3- to 0.5-cm pieces (Fig. 3B) and further processed according to the TS-MAP workflow. We could verify comprehensive and distinct cell-surface labeling of intact patient tumors by immunofluorescence imaging of tissue sections and FACS analysis of tumor SCSs (Fig. 3 B and C). We found a strikingly heterogeneous SURFME protein expression between patient tumors, and DEP analysis depicted a unique pattern of overexpressed proteins apparently unrelated to histology or grade (Fig. 3D). Hotspots of highly expressed SURFME proteins diverged according to subcategory, with a potential enrichment of multipass transmembrane proteins in oligodendrogliomas (ODGs) as compared with astrocytomas (Fig. 3E). We selected EGFR, BCAN, CD81, TF, and MCT2 for further studies based on their divergent, normalized relative abundances between high- and low-grade tumors according to TS-MAP/SURFME data (Fig. 3F). Another criterion for our candidate selection was that EGFR and TF are currently explored as ADC targets in clinical trials, and previous studies pointed at their potential as treatment targets in cancer (26–31). We next performed immunofluorescence stainings of available matched patient tumors for EGFR, TF, BCAN, CD81, and MCT2. In accordance with the TS-MAP profiling data, we found a clear overexpression of EGFR and TF in GBM as compared with ODG (Fig. 3G). Moreover, BCAN was highly expressed in patient tissue #5GBM as compared to #8ODG, and both MCT2 and CD81 were strongly expressed in #1GBM vs. #8ODG patient tissue (Fig. 3G). Importantly, as part of the validation, the selected proteins were preferentially expressed at the cell surface in tumor sections.

Solid tumors, including gliomas, are characterized by the infiltration of a variety of stromal and immune cells that contribute to the malignant process (32). In particular, reprogrammed vascular cells and immunosuppressive tumor-associated macrophages (TAMs) are prevalent in GBM, and their eradication has emerged as a potential treatment strategy (33, 34). To understand whether the TS-MAP approach included these GBM-microenvironment subsets, we performed immunofluorescence costaining for cell-surface biotinylation and specific markers of immune cells (CD45⁺), TAMs (CD68⁺), endothelial cells (CD31⁺), and EGFR⁺ cells (SI Appendix, Fig. S4). We consistently found a clear biotinylation labeling of these cell entities in tissue sections (SI Appendix, Fig. S4A), and confocal microscopy of tumor SCSs could confirm distinct surface biotinylation of CD45⁺, CD68⁺, CD31⁺, and EGFR⁺ tumor cells (SI Appendix, Fig. S4B). However, not only tumor resident but also systemic immune cells are of interest in cancer. Therefore, we isolated peripheral blood mononuclear cells (PBMCs; CD45⁺) from GBM patient blood for surface biotinylation and further enrichment of CD14⁺ monocytes. FACS analysis revealed comprehensive labeling of PBMCs and monocytes (90 to 95% of total) (SI Appendix, Fig. S4C), and specific cell-surface labeling was visualized by confocal microscopy (SI Appendix, Fig. S4D). Together, we conclude that patient tumors exhibit a remarkable TS heterogeneity that apparently is not associated with grade and histology, which reinforces the need for techniques allowing individualized target identification. TS-MAP may fulfill this need by providing a platform for direct surfaceome profiling of tumors and, in parallel, specific subsets of peripheral cells that could be used for target selection, but also improved diagnosis and monitoring.

We finally wanted to explore the possibility of maintaining surfaceome-labeled patient glioma specimens under endocytosis permissive conditions to profile the endocytome of intact tumor tissue. After extensive method optimization, we chose 1.5 h as a good time point of constitutive endocytosis, as this yielded a clearly visible, intracellular signal in tumor sections (Fig. 4A). Importantly, we could confirm efficient internalization into vesicular structures as well as eradication of remaining surface biotinylation by MesNa, using high-resolution imaging (Fig. 4B), FACS (Fig. 4C), and Western blotting (Fig. 4D). As further evidence of endocytic internalization in this setting, the intracellular biotinylation signal showed colocalization and time-dependent transition from early endosomes (EEA1⁺) to late endolysosomes (CD63⁺) (Fig. 4E). From these data, we were encouraged to profile the surfaceome and endocytome from a patient with recurrent GBM (#10GBMr). LC-MS/MS analysis and SURFME filtering identified 348 surface proteins, out of which 299 were also endocytosed (Fig. 4F). We found an additional 49 proteins expressed exclusively at the surface and 15 confined to the endocytosed fraction (Fig. 4F). We argued that proteins that are strongly expressed at the surface and efficiently endocytosed represent targets with a potential application for ADC or RIT. By ranking identified SURFME proteins according to their relative abundance in the surfaceome and endocytome, we selected identities with high-abundance ranking in both fractions (Fig. 4 G, Top Right quadrant) for analysis of matched tissue sections (see SI Appendix, Table S2 for a detailed abundance ranking of surfaceome and endocytome proteins). Using high-resolution confocal microscopy, we could observe a strong surface and intracellular signal of EGFR, BCAN, and TF in #10GBMr, thus corroborating the LC-MS/MS data (SI Appendix, Fig. S5). Altogether, these results provide proof of concept of the TS-MAP platform as a versatile tool to profile the highly diverse tumor surfaceome and functional endocytome in patient tumors.

U3065-MG and U3082-MG cells were obtained from the HGCC biobank, Uppsala University, Uppsala, Sweden (14). Cells were seeded on dishes precoated with 10 μg/mL poly-L-ornithine (Sigma-Aldrich, P3655) and 10 μg/mL laminin from Engelbreth-Holm-Swarm murine sarcoma basement membrane (Sigma-Aldrich, L2020) and routinely cultured in primary cell medium, composed of Neurobasal (Gibco, 21103-049) and Dulbecco's modified Eagle medium/F12 (Gibco, 31331-028) media (1:1 mix) without serum, supplemented with 1× N2 (Gibco, 17502-048), 1× B27 (Gibco, 12587-010), 10 ng/mL human recombinant FGF2 (Peprotech, 100-18B), 10 ng/mL EGF (Peprotech, AF-100-15), 100 U/mL penicillin and 100 μg/mL streptomycin (PEST; Sigma-Aldrich, P0781). For 3D cultures, GBM cells were seeded at 2 × 10⁵ cells/mL in uncoated dishes, in the same culture medium as described above, and cultured on an orbital shaker at 90 rpm. At 7 to 10 d, spheroids reached a diameter of 0.3 to 0.5 mm, and the presence of a hypoxic core was observed using Image-iT Green Hypoxia Reagent (Thermo Fisher Scientific, I14833). Spheroid and 2D cultures were dissociated using TrypLE express (Gibco, 12605010). All cells were grown in a humidified 5% CO₂ incubator at 37 °C.

Clinical specimens were collected from patients referred to the Neurosurgery Department at Lund University Hospital. Inclusion criteria were age 18 y or above, WHO performance status 0 to 2, and ability to give written informed consent. The study was carried out according to the ICH/GCP guidelines, in agreement with the Helsinki declaration, and approved by the local ethics committee, Lund University (Dnr. 2018/37). Patients were diagnosed by routine MRI of the brain and surgical and pathological procedures, received standard oncological treatment, and were followed up according to local and national recommendations. Clinical-pathological characteristics of included patients are given in SI Appendix, Table S1. Fresh samples of macroscopically viable tumors were directly processed for TS-MAP or cryopreserved by snap freezing in isopentane for further immunofluorescence (IF) evaluation (see below).

For surfaceome profiling in 3D and fresh patient samples, we developed a procedure emanating from studies with HeLa cell-line monolayers (49). Subconfluent primary GBM cell 2D monolayers, 0.3- to 0.5-cm diameter spheroids and mouse or patient tumor pieces (0.3 to 0.5 cm) were preincubated on ice for 10 min and maintained on ice to prevent internalization (under 100 rpm orbital agitation for spheroids and tissue samples). Cells and tissues were washed twice with ice-cold phosphate-buffered saline (PBS) containing MgCl₂ and CaCl₂ (Mg/Ca–PBS; Thermo Scientific) adjusted to pH 8.0, followed by incubation with 1 mg/mL of a membrane impermeable and cleavable biotin moiety (EZ-Link Sulfo-N-hydroxysuccinimide-SS-biotin, Thermo Scientific, 21331) in Mg/Ca–PBS, pH 8.0 for 30 min protected from light, and washed with ice-cold Mg/Ca–PBS. To preclude unspecific cellular entry of the biotinylation reagent or of secreted/extracellular proteins, free biotin was quenched with 0.1 M glycine in Mg/Ca–PBS for 10 min, and cells were subsequently washed twice with ice-cold Mg/Ca–PBS for further processing.

For endocytome profiling, following cell-surface biotinylation, endocytosis was initiated by the addition of prewarmed primary cell medium for the indicated time periods in a humidified 5% CO₂ incubator at 37 °C. Endocytosis was finalized by incubation on ice for 10 min, and remaining cell-surface biotin was removed with the membrane impermeable, reducing agent, MesNa (200 mM, sodium-2-mercaptoethanesulfonate; Thermo Scientific, M1511), dissolved in 50 mM Tris pH 8.6 containing 100 mM NaCl, 1 mM ethylenediaminetetraacetic acid (EDTA) and 0.2% bovine serum albumin (BSA) twice for 15 min each time at 4 °C in the dark. Samples were then washed with ice-cold Mg/Ca–PBS and incubated with iodoacetamide (5 mg/mL, Sigma-Aldrich, I6125) in Mg/Ca–PBS for 10 min in the dark, and subsequently washed with Mg/Ca–PBS. For LC-MS/MS experiments, an additional cell-surface biotin blocking step was performed prior to cell lysis using unconjugated streptavidin (50 μg/mL; Sigma-Aldrich, S4762) diluted in Mg/Ca–PBS containing 1% BSA for 30 min at 4 °C, followed by extensive washes with 12 mL of PBS 0.1% Triton X-100, followed by 10 mL of radioimmunoprecipitation assay (RIPA)/PBS 0.1% Triton X-100 1M NaCl 1:1 (vol/vol), and finally 10 mL of PBS 0.1% Triton X-100 to remove unbound streptavidin. As a control of cell-surface biotin removal efficiency, we verified the absence of signal after the MesNa and streptavidin blocking steps in surface-biotinylated cells that did not undergo endocytosis.

Biotinylated surfaceome and endocytome (1.5-h endocytosis) from 2D, spheroid, mouse, and patient samples, as indicated, were quantified on a per-cell basis after gentle detachment/dissociation with TrypLE express (Thermo Fisher Scientific). Endocytosis samples were permeabilized for 30 min with 0.5% saponin in PBS. Unspecific binding was blocked with 3% BSA in PBS for 30 min at room temperature (RT). Biotinylated cell labeling was performed with 5 μg/mL streptavidin-Alexa Fluor (AF)-488 (Life Technologies, S32354) in Mg/Ca–PBS containing 3% BSA for 30 min at 4 °C. Data were acquired on an Accuri C6 flow cytometer and analyzed using FlowJo software (version 10). Results are expressed as normalized mean fluorescence intensity (MFI) after subtraction of the values of negative control cells.

Biotinylated spheroids and mouse and patient samples were fixed for 15 min in 4% paraformaldehyde (PFA), cryopreserved by overnight incubation in 0.5 M sucrose, and embedded in optimal cutting temperature (OCT) medium (Fisher Scientific, 12678646) for cryosectioning. Alternatively, biotinylated tissue samples were enzymatically dissociated with TrypLE express and DNase I (Thermo Fisher Scientific, 90083), for 10 min at 37 °C to obtain SCSs, which were filtered through 70- and 40-µm nylon cell strainers, and red blood cells (RBCs) were removed using RBC Lysis Buffer (BioLegend, 420301). Biotinylated primary GBM cells growing as 2D monolayer and SCSs from spheroids and tissues were seeded in eight-well chamber slides (Ibidi, 80827), washed with PBS, fixed with 4% PFA for 10 min, and permeabilized for 30 min with 0.5% saponin diluted in PBS. Nonspecific sites were blocked with PBS containing 3% BSA. Cryosections were blocked in PBS with 5% goat serum (Dako, X0907) for 1 h at RT. Biotinylation was visualized by labeling with 5 µg/mL streptavidin-AF-488 (Thermo Fisher Scientific, S32354) in PBS-3% BSA or PBS-5% goat serum for 30 min at 4 °C. IF staining of cells and sections was performed overnight at 4 °C using the following primary antibodies: Rabbit anti-EEA1 (1/200, Abcam, ab2900), rabbit anti-EGFR (1/200, Abcam, ab32198), rabbit anti-CSPG5 (1/200, Thermo Fisher Scientific, PA5-75459), mouse anti-CD63 (1/100, Abcam, ab8219), mouse anti-CD45-PE (1/25, BD Biosciences, 555483), rabbit anti-CD68 (1/800, Cell Signaling Technology, D4B9C), mouse anti-CD31 (1/100, Dako, JC70A), mouse anti-CD44 (1/100, Dako, DF1485), mouse anti-CD14-PE (1/25, BioLegend), rabbit anti-BCAN (1/200, Abcam, ab111719), rabbit anti-CD81 (1/200, Abcam, ab219209), rabbit anti-TF (kindly provided by Wolfram Ruf, Center for Thrombosis and Hemostasis, University of Mainz, Mainz, Germany), or rabbit anti-MCT2 (1/100, Thermo Fisher Scientific, PA5-112712). After washing, cells were incubated with anti-mouse or anti-rabbit AF-488 or AF-546-conjugated secondary antibody (1/500) (Thermo Fisher Scientific, A11001, A11060, A11008, and A11010) for 1 h at 4 °C. Nuclei were counterstained with Hoechst 33342 (1/20,000p) (Thermo Fisher Scientific, 1399) for 10 min. Sections were finally mounted with Fluorescence Mounting Medium (Dako, S3023). In some cases, subconfluent GBM cells were transfected with a plasmid (250 ng) encoding CD63-mCherry (kindly provided by Jennifer Lippincott-Schwartz, Howard Hughes Medical Institute, Ashburn, VA) using Lipofectamine 2000 (Thermo Fisher Scientific) according to manufacturer protocol.

Imaging (fixed and live cells) was performed using a Zeiss LSM 710 confocal microscope equipped with a 34-channel confocal spectral detector or Airyscan array detector unit. Light sources were a diode laser 405–SF30 (405 nm), a Lasos LGK 7812 argon laser (458 nm, 488 nm, and 514 nm), and a HeNe633 laser (633 nm). Plan-Apochromat 20×/0.8, or Plan-Apochromat 40×/1.40 differential interference contrast (DIC) M27 oil immersion or Plan-Apochromat 63×/1.40 DIC M27 oil immersion objectives were used. Images were processed using ZEN 2.1 black edition (Carl Zeiss).

Biotinylated samples were lysed for 20 min at 4 °C in RIPA buffer (150 mM NaCl, 50 mM Tris pH 8.6, 1% Nonidet P-40, 0.1% sodium dodecyl sulfate, 0.05% sodium deoxycholate) containing 2× protease inhibitors (Roche Diagnostics, 04693124001). Lysates were clarified by centrifugation at 18,000 × g for 10 min at 4 °C. The soluble fraction was collected, and total protein was quantified using BCA Protein assay kit (Pierce, 23225). Lysates were diluted 1:4 with Mg/Ca–PBS supplemented with protease inhibitors, filtered with a 0.45-µm surfactant-free cellulose acetate syringe filter, and then applied to a HiTrap streptavidin HP-1 mL column (GE Healthcare, 17-5112-01) pre-equilibrated in PBS 0.1% Triton X-100 using a peristaltic pump set at a flow rate of 250 μL/min. The column was then transferred to an high-pressure liquid chromatography (HPLC) UPC 900 system (Amersham Biosciences) equipped with an online ultraviolet (UV) detector set at 280 nm, and washed with 10 mL of PBS 0.1% Triton X-100, followed by 10 mL of RIPA/PBS 0.1% Triton X-100 1 M NaCl 1:1 (vol/vol), and finally 10 mL of PBS 0.1% Triton X-100 to remove nonbiotinylated proteins at a flow rate of 1 mL/min. Biotinylated proteins were then eluted from the column by reduction of the protein-SS-biotin linker with 10 mL freshly prepared 150 mM MesNa in PBS 0.1% Triton X-100 applied at a reduced flow rate (125 μL/min). One volume of 20% trichloroacetic acid was added to the collected eluate to precipitate proteins by incubation for 30 min on ice and centrifugation for 10 min at 18,000 × g. Protein pellets were finally washed twice with 2% sodium acetate and resuspended in 6 M urea buffer for LC-MS/MS sample preparation.

Protein pellets, isolated as described above, were dissolved in 50 mM ammonium bicarbonate (AMBIC) buffer containing 6 M urea (5× the sample volume). Disulfide bonds within protein samples were then reduced by incubating samples with 10 mM dithiothreitol for 1 h at 56 °C at 300 rpm and alkylation of free sulfhydryl groups was done with 30-min incubation with 20 mM iodoacetamide in the dark at RT, followed by buffer exchange with 50 mM AMBIC, pH 8.0. Protein samples were digested with 0.4 μg/μL of sequencing grade porcine trypsin (Promega, V511A) in AMBIC at an enzyme-to-protein relation of ∼1:100 overnight at 37 °C under 350 rpm agitation. The digestion was stopped by adding 1.73% trifluoroacetic acid (TFA) and samples were dried using a Speed Vac, resuspended in 100 µL of 0.1% TFA in dH₂0 and desalted with Ultra MicroSpin silica C18 columns (The Nest Group, SUM SS18V). Samples are eluted with a 50% acetonitrile (ACN), 0.1% TFA solution in dH2O by spinning at 200 × g for 1 min and then dried with the Speed Vac and stored at −80 °C until further analysis.

LC-MS/MS analysis was performed on a Tribrid Mass Spectrometer Fusion (Thermo Fisher Scientific) equipped with a Nanospray Flex ion source and coupled with an EASY-nLC 1000 Ultra high pressure liquid chromatography (UHPLC) pump (Thermo Fisher Scientific). For the analysis, samples were resuspended in 2% ACN-0.1% TFA, and peptide concentration was adjusted to 0.5 µg/µL by measuring 215 nm absorbance using a DeNovix DS-11 spectrophotometer, and 1 µg of peptides was injected in one to three independent runs depending on total sample peptide amount. Peptides were concentrated on an Acclaim PepMap 100 C18 precolumn (75 μm × 2 cm, Thermo Fisher Scientific) and then separated on an Acclaim PepMap RSLC column (75 μm × 25 cm, nanoViper, C18, 2 μm, 100 Å) set at 45 °C and a flow rate of 300 nL/min. Solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile) were used to create a nonlinear elution gradient. The Orbitrap Fusion was operated in the positive data-dependent acquisition (DDA) mode. The peptides were introduced into the LC-MS via a stainless steel nano-bore emitter (outer diameter [OD] 150 µm, inner diameter [ID] 30 µm) with the spray voltage of 2 kV and the capillary temperature set at 275 °C. Full MS survey scans from m/z 350 to 1,350 with a resolution of 120,000 were performed in the Orbitrap detector. The automatic gain control (AGC) target was set to 4 × 10⁵ with an injection time of 50 ms. The most intense ions (up to 20) with charge states 2 to 5 from the full scan MS were selected for fragmentation in the Orbitrap. The MS2 precursors were isolated with a quadrupole mass filter set to a width of 1.2 m/z. Precursors were fragmented by high-energy collision dissociation (HCD) at a normalized collision energy (NCE) of 30%. The resolution was fixed at 30,000 and for the MS/MS scans, the values for the AGC target and injection time were 5 × 10⁴ and 54 ms, respectively. The duration of dynamic exclusion was set to 45 s and the mass tolerance window was 10 ppm. For comparison, some samples were analyzed on a Thermo Easy nLC 1000 system (Thermo Fisher Scientific) coupled online to a Q-Exactive HF-X mass spectrometer (Thermo Scientific). Peptides (1 µg) were loaded on a precolumn (Thermo Scientific; ID 75 μm × 2 cm, temperature 35 °C) and then separated on an EASY-Spray column (Thermo Scientific; ID 75 μm × 25 cm, temperature 45 °C). A nonlinear gradient of buffer B (80% acetonitrile, 0.1% formic acid) in buffer A (aqueous 0.1% formic acid) was applied at a flow rate of 300 nL/min. One full MS scan (resolution 120,000 @ 200 m/z; mass range 375 to 1,500 m/z) (resolution 15,000 @ 200 m/z). The precursor ions were isolated with 1.2 m/z isolation width and fragmented using higher-energy collisional-induced dissociation at a normalized collision energy of 28. Charge state screening was enabled, and singly charged ions as well as precursors with a charge state above 6 were rejected. The dynamic exclusion window was set to 40 s. The automatic gain control was set to 3 × 10⁶ for MS and 1 × 10⁵ for MS/MS with ion accumulation times of 50 ms and 19 ms, respectively. The intensity threshold for precursor ion selection was set to 5.3 × 10⁵.

The raw DDA data were analyzed with Proteome Discoverer 2.3 (PD 2.3) Software (Thermo Fisher Scientific). Peptides were identified using SEQUEST HT against UniProtKB human database (release 2020_05), using the following parameters: Static modification, cysteine carbamidomethylation and dynamic modifications, N-terminal acetylation and methionine oxidation. Precursor tolerance was set to 10 ppm and fragment tolerance to 0.05 ppm. Up to two missed cleavages were allowed and Percolator was used for peptide validation at a q value of maximum 0.05. Protein abundance was calculated based on intensity of unique peptides after low abundance resampling missing value imputation and total peptide amount normalization.

An in-house list of curated protein IDs belonging to the human surfaceome was generated. An initial set of plasma membrane–associated proteins was constructed by merging nine terms from public databases: 2,886 proteins predicted by SURFY (retrieved from http://wlab.ethz.ch/surfaceome/) (1); 5,196 listed in plasma membrane GO term (GO:0005886); 935 in cell surface GO term (GO:0009986); 437 in GO’s external side of plasma membrane term (GO:0009897); 939 reviewed UniProt GPI-anchored proteins; 2,356 reviewed UniProt single-pass transmembrane proteins; 2,819 reviewed UniProt multipass transmembrane proteins; 3,724 reviewed UniProt cell membrane proteins; and 2,726 reviewed UniProt proteins within extracellular domain topological term. Duplicates and false positives were filtered out by curation based on further database annotation (from UniProt, GO, and GeneCards) and literature review (PubMed.gov). We finally identified a total of 3,317 proteins, hereafter defined as SURFME (see also Results section and Dataset S1).

Human homologs to mouse symbols were obtained with the BiomaRT package. The 135 mouse symbols not mapped to human equivalents by BiomaRT were manually searched using mouse genome informatics homology information (http://www.informatics.jax.org/). Bioinformatics analyses were conducted in R version 4.0.4 and figures were generated using the packages ggplot2, RColorBrewer, viridis, VennDiagram, venneuler (in combination with http://bioinformatics.psb.ugent.be/cgi-bin/liste/Venn/calculate_venn.htpl to perform six-group overlap comparison), and pheatmap (clustering method used was “ward.D2”). GraphPad Prism 8.3.1 was used to generate some of the bar plots, and graphical illustrations were created with BioRender.com. Figure design and composition were performed with Adobe Illustrator 25.2. Statistical analyses were performed in GraphPad Prism using two-way ANOVA with Tukey’s post hoc test for multiple comparison. All values with p < 0.05 were considered statistically significant.